Serious' cheap OB

Discussion in 'Speakers' started by Serious, Oct 18, 2016.

  1. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Daily update :p:
    Realized that I fucked up because my simulation was off. I will have to redo the measurements for the simulation. Maybe the impedance measurement was off. I quickly pulled some shit out of my ass and increased the zobel cap to 47uF. Looks (and sounds) much better now. Better tonality and deeper soundstage. Also much better off-axis response when standing up. Roll-off should probably be at a slightly lower frequency and more gentle. I won't reverse the bass driver for now.

    This is what it looks like without the usual B&K target compensation now. Right speaker. The right Fostex driver seems to extend a little better. This isn't super accurate - the measurements weren't taken at exactly the same spot, but it should give you a rough idea.
    Xover 7.png

    I know I'm getting there eventually...

    EDIT:
    Also always helps to have the HD800 on hand for a quick reality check. These drivers are obviously less capable than the HD800, but as it is right now I also prefer the tonality of my modded HD800. My HD800 isn't all that much brighter, just mainly more extended (towards both ends). Midrange is more realistic on the HD800.
     
    Last edited: Oct 26, 2016
  2. Cspirou

    Cspirou They call me Sparky

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    4,411
    Dislikes Received:
    5
    Trophy Points:
    113
    Location:
    Northwest France
    What are you using to simulate this?
     
  3. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Actually using the german "WinBoxSimu" software. I think there are many others that do essentially the same thing. I use that because I still had it on my laptop.
     
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Thought I'd share some more measurements (and impressions). I tried a few different crossover implementations, but ultimately settled on the 3.9mH / 3.4Ohm, 47uF one again. The crossover isn't perfect, but it's almost impossible to make it perfect with the current baffle and driver choice. Would really need a bigger midrange and perhaps a bigger baffle to be able to make a real textbook first order crossover design. What I have right now is also not a real 1st order; it's steeper than that in reality. Still, I can play both drivers in positive acoustic polarity and I don't get a deep null, so that's good (meaning that it's not a 2nd order, either). I do, however, get a suckout at about 500Hz above the listening position.

    Here are some measurements that I found interesting:

    Personally I prefer to look at the step response, but here's a 300Hz square wave at the listening position. Some people like the square waves.
    300Hz.jpg
    As you can see, the first one has a good shape indicating good phase and frequency response, but then the reflections start to come in and mess up the whole picture. I find this highlights where the more traditional frequency response measurements fail in my opinion: Music has a lot of transient information. Some of the room effects may need some time to really come into play.

    Here are the CSDs at the listening position, or actually a little closer to the drivers than the actual listening position, because the couch was too reflective. Windowed to 5ms (BH4 window I believe). The crossover is at around 500Hz. I intentionally let the CSDs go down to 347Hz to capture the xover region.
    Note these are only a 20db range, compared to the 35db headphone range. This already hits the noise floor.
    R ca 1m50 CSD 347Hz.png
    (When standing up, there's a suckout at 500Hz and when sitting lower there's too much energy there)
    This is from the right speaker. The FE83En in the right speaker seems to extend better than the left driver. Extension isn't an issue with the right one. It's actually possible that this got better with some burn-in. I'm not sure if the measurements show this, but I do hear it as more extended now. As expected, there's some ringing at 6 and 9kHz, but these banana paper drivers present the ringing so differently from other drivers. It doesn't bother me, but the ringing does hurt the treble delineation and clarity. The deep notch between 2-3kHz is actually slight ringing.

    I wonder how much brighter the 4" Fostex sounds since @Hooncake complained about brightness. Or maybe that's the BLH with small baffles.

    Another measurement that I find useful for speaker measurements: The impulse response over a longer time-scale with a log Y-axis.
    R IR 75ms.png
    Mostly just nice, diffuse decay. I think @Hrodulf would be happy seeing this, but would probably want even less reverb. The T60 is a little high at about 300ms for me. I prefer to keep it between 0.2 and 0.25s across the band, or maybe even lower.
    I also attached the RT60 plot. This is something really awesome that REW does. If this rises towards higher frequencies, it will sound brighter than FR measurements would suggest. It's mostly just a good idea to keep it flat looking, although I would prefer less reverb. Honestly I really like listening to speakers outdoors.
    I don't really have any absorption in my room. The only things I did were some first reflection point side-wall absorption to help with the imaging since my room is very narrow. A thin blanket and sheepskin on the left wall and a thick pillow at widebander height on the right wall.

    What I find the FE83En to lack most is the resolution and clarity I need. It almost feels like an ambient information removal filter is applied to the music compared to the HD800, but I've not heard one speaker that managed to reproduce all the ambient information in HFO that I can hear with the HD800.

    The Beta 15A feels very snappy, but a little less controlled than ideal for my tastes. At about 90db at the listening position (which is closer to 95db at 1m), it doesn't go over 1% distortion until below 30Hz, which is actuallly pretty awesome for a speaker. I actually looked at the distortion plots for speakers measured by Stereoplay and it turned out that almost all of them went above 1% at 90db in the bass region. Even speakers that cost more than a car. Or even the new B&W 800D3 with its dual 10" bass drivers. It's still possible that my mic sensitivity isn't calibrated perfectly, but I doubt it since my measured sensitivity at 2.83V/8Ohm matches the specs very well.
    When I go for a more efficient mid/high solution I will have to double up on woofers. Not sure how much the Ragnarok would like the 4Ohm load throughout the bass region, but I doubt it's a big issue. However it could become a bigger issue if I played the midrange all the way down without a high-pass like I do right now, essentially bringing the impedance down to about 2.5 Ohms with two woofers. Even then, the efficiency would be around 96db, so might not be a huge issue.

    Overall I like the sound of a widebander with no xover parts, but I don't think I'd be ok with the bass quality of the traditional BLH approach, even with 8" drivers or so. The 15" in an open baffle is just a whole different ballgame in terms of bass-quality.

    BTW: This looks pretty awesome: http://www.pureaudioproject.com/product/trio15-voxativ-open-baffle-speakers/
    Now if it could only be time-coherent :p. Also the woofers seem to be modified Alpha 15As. Not sure how I feel about that. Also midrange is centered on the baffle. This is almost a requirement for a commercial speaker, but I would avoid this if it was my own speaker (after all I did). Remember when @Marvey complained about 3db 1.5kHz shout with the Cicadas on the OB? That was because the driver was centered on the baffle.
     

    Attached Files:

    Last edited: Oct 29, 2016
  5. Hooncake

    Hooncake Mid-Fi Purgatory Redemption

    Friend
    Joined:
    Dec 11, 2015
    Likes Received:
    691
    Dislikes Received:
    6
    Trophy Points:
    93
    Location:
    NYC
    Haha this post is way above my head. I am a simple speaker noob.

    Hopefully I didnt complain too much about the 103en highs. I think they are fine and I can listen to them for a long time. They are brighter than HD650 and OS though. The most accurate way I can describe is that the midrange and higher frequency seem to have similar amount of volume.

    I was more complaining about OSMT's high that when I nearly filled in the entire cabinet with Acousta-fluffy-cotton-stuff the highs became prickly or spiky. Sibilant?

    We will have veteran NYC people come over for listen in the near future and will post more impressions.
     
    Serious likes this.
  6. OJneg

    OJneg The Most Insufferable

    Friend BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,545
    Dislikes Received:
    13
    Trophy Points:
    113
    Location:
    Grand Rapids
    Get with da program, all the cool kids have moved on to horns :cool:
     
  7. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    I actually thought about horn designs recently. Interested in the more controlled dispersion, but it does come with issues. They'd need to be really big, which I'm fine with, but that also limits how close you can put the drivers. Also because of horn phase shifts a time coherent full-range passive horn speaker (meaning using horns for all the drivers, except maybe the bass) most certainly won't be happening. Using a horn only for the tweeter is not something I'd plan to do.
    Besides, the bigger full range drivers are more efficient than many small horns anyway.
     
  8. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Tried a new crossover approach today. I assumed that the FE83En would roll off 6db per octave starting at about 600Hz or so, -3db point at 400Hz or so based on the driver LF extension and my baffle simulation. Makes sense, right? I then tried to get the woofer as close to an acoustical 1st order low pass starting at 400Hz or so. For this I measured the driver at about 1m distance to get less of the room effects.
    I ended up with the 2nd coil added again - I think this is makes it about 5.5 to 6mH (the 2nd big coil I have doesn't have a number on it) - and I increased the zobel resistor to 5.5 Ohm, still with the 47uF cap (actually 3 caps in parallel). The result? Much better driver integration. I can't believe I didn't try this before. It was so simple - anyone could design a crossover like this. Widebander with 1st order baffle high pass + Woofer with 1st order low pass = win.

    Now a D'Appolito design with one of the more resolving widebanders would be interesting, but mirroring the whole thing would make the speaker nearly 6ft tall. That would actually be enough space for 4 woofers per speaker. Heh, 4 15" woofers in series-parallel on two sloped baffles, so that they're time-aligned at the listening position would probably be pretty awesome.

    Here's the step response at about 1m50, just in front of the couch:
    new step.png
    I think the dip around 90ms after the peak is the Fostex driver itself and doesn't reflect the driver integration. The initial peak now isn't as high as it should be because of the slightly warm FR. This is a good thing (see below).

    And here's the uncompensated FR at the same position (B&K compensated one is attached):
    R FR new ca 1m50.png
    The crossover should be around 500Hz now (It was slightly higher than that before).


    EDIT: Midrange tonality seems quite nice now. Actually reminds me a little of my HD800. Maybe slightly more 2kHz and slightly less 3-5kHz versus my HD800, but nothing major I'd say. I'd say the midrange tonality is better than Utopia. The Fostex drivers does have some midrange shout.
     

    Attached Files:

    Last edited: Oct 30, 2016
    purr1n likes this.
  9. Cspirou

    Cspirou They call me Sparky

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    4,411
    Dislikes Received:
    5
    Trophy Points:
    113
    Location:
    Northwest France
    What sort of amplification are you using? Is it bi-amped?
     
  10. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    I use my Ragnarok. Efficiency is about 90db/2.83Vrms. Impedance should be about 4-5 Ohms in the bass and lower mids (from the parallel drivers) and 8 Ohms from the crossover on up. I actually use the Ragnarok in mid-gain most of the time.

    I'm sure most people could use bigger woofer baffles for a bit more bass. It's a bit lean the way it is now. The other option would be a more efficient woofer.
     
  11. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    45,217
    Dislikes Received:
    70
    Trophy Points:
    113
    Location:
    Antarctica
    I don't remember your design, but instead of MTM you could locate the driver's far from each other instead. MTM comes with its own type of weirdness. There is no perfect solution.

    On step response, looks like the wide bander is in front of woofer. You could slope the OB back or mount the wide bander behind the baffle and chamfer the panel in front. I don't think perfect step responses really matter that much anyway.
     
  12. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    45,217
    Dislikes Received:
    70
    Trophy Points:
    113
    Location:
    Antarctica
    Consider a U frame if you want lower bass extension. Can add stuffing behind too. The trade off is slower transient response.

    You can experiment easily by taping pieces of cardboard.
     
  13. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Why would I want to seperate the drivers more than necessary? I think that an MTM should work reasonably well for such a low crossover frequency, but most of those designs seem to be the typical 2kHz crossover, where the drivers will usually be too far apart for the proper dispersion.
    I think that the imaging will get worse with more distance between the drivers.
    What sort of weirdness? The way I see it, instead of a bump and a dip, you get a dip on both sides.

    I tried to tape a piece of cardboard and I thought it made the bass sound more like a closed box, so I removed it again. I guess with some more experimentation I could improve it, but then I should probably also look into improving the crossover.

    It does look like it, but it's not as simple. I'd basically have to redesign the entire crossover. The widebander is on its own baffle about 4" behind the woofer baffle. Right now the drivers should have the same distance from the listening position. The signals do add at the crossover frequency, just not quite 6db afaik.
    What's the frequency for that dip - 90ms after the peak of the IR? I'm not sure if it's in the crossover region.


    Actually, let me get some more measurements of the crossover. It's not so easy because the floor reflections at the listening distance seem to be at a similar frequency as the crossover region. I consider this project mostly done. Yes, it could use some polishing, but I'd rather put that energy into an OB with a more capable widebander. I do like the 1st order crossover approach with a relatively high XO frequency and absolutely no parts for the widebander.
     
    Last edited: Dec 26, 2016
  14. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    Here's a measurement (at the listening position) I took really quick:
    FR:
    FR.png
    As you can see, it's not perfect, but they do add. As I said, it's not easy to get an accurate measurement without an anechoic chamber.

    Here's the FE83En IR and step response. I think this is what causes the dip in the step response. I think that's the dip at 2kHz in the FR. FE83En IR.png

    Here's the same thing, but with both drivers:
    Sum IR.png

    Note: The FR was without compensation, but my target has actually changed a little - the lower mids and bass for music are mostly mono information.
     
  15. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    45,217
    Dislikes Received:
    70
    Trophy Points:
    113
    Location:
    Antarctica
    Exactly. The dip on both sides surrounding a more narrow vertical sweet spot can be more annoying and than wider dip and bump. Depends upon speaker placement, driver height, and listening position.

    I find myself usually sitting up on the couch or slumped over drooling, so not a fan of MTMs. Then again, my last traditional multi way speaker was an MTM.
     
  16. Aklegal

    Aklegal Facebook Friend

    Contributor
    Joined:
    Mar 1, 2016
    Likes Received:
    91
    Dislikes Received:
    0
    Trophy Points:
    28
    I typically build OB speakers from kits (in the middle of building my third) so I am a bit of a noob when it come to crossover design etc, but any reason why you can't reduce the Fostex baffle width? I was wondering if this would make any difference since you are not using a tweeter. So again pardon the noob question.

    Have you considered an H Frame instead of a U?
     
  17. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    I usually keep my head at the same height for listening, so this wouldn't be an issue. Less height would actually be good. I think most manufacturers design their speakers for a higher listening position than I would. The MTM OB with the Voxativ widebander has its widebander pretty low, though (only 55cm from the floor I believe).

    Since I'm using the baffle as my high pass for the Fostex driver, the baffle dimensions will influence the crossover. It just depends on the crossover frequency you're aiming at. You can use Edge for simulating a flat baffle response: http://www.tolvan.com/edge/help.htm

    Right now it's a flat baffle and not a U-Frame. Maybe I'll experiment some more with H-Frames. What's the easiest way to simulate a driver + baffle response for an H frame?
     
    Last edited: Dec 26, 2016
  18. Aklegal

    Aklegal Facebook Friend

    Contributor
    Joined:
    Mar 1, 2016
    Likes Received:
    91
    Dislikes Received:
    0
    Trophy Points:
    28
    I gotcha, makes sense. So the wider baffle allows you to design a less complicated crossover (assuming you use one for this design).

    You are way above my head on this one since I have never come close to designing a speaker. :(
    My layman's answer is that I have read alot of comparisons between H and U frames for better bass response and while obviously not conclusive many point to H frames as having better bass response but I imagine there is give and take there as well since not everyone selects the same woofer. I did get the general impression that U frames are often chosen for aesthetic purposes (They are way more spouse friendly).

    I guess my original question was based upon you separating the Fostex baffle from the woofer baffle and thinking that then you could do whatever you wanted for the woofer section. Now as call in's say on a local radio show I listen to often "I'll hang up and listen"
     
  19. Serious

    Serious Inquisitive Frequency Response Plot

    Friend BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    1,495
    Dislikes Received:
    0
    Trophy Points:
    113
    Location:
    Munich
    As I was looking through my stuff I found in-ear measurements from about a month ago. I thought they were very interesting:

    First up, here's the FR with both speakers playing with a microphone in my ear (can't remember which one):
    InEarBoth.png
    I wonder where the dip around 1kHz is coming from. I don't think I can hear it. The rest however looks similar to how I hear it: Bass extends to 40Hz, midrange is generally pretty smooth and the treble is rolled off. I know @ultrabike would freak out about the lacking outer ear gain, but I think a lot of this is a speaker off-axis FR thing. I don't think the upper mids are too low at all. The lack of treble was surprising to me, but subjectively it made sense. EQing everything above 5kHz up 5db does the trick*. To be honest I think this is the most accurate FR I've seen of those speakers. (I think it's best to ignore the dips in the treble. It's an in-ear measurement after all.)
    Those measurements also tell you how much the perceived FR depends on the off-axis response. I damped the side-walls which is why not a lot of treble gets reflected off there and why it's warmer than the omni mic measurements would suggest.
    EQing the treble up makes it sound much closer to my HD800 in tonality (which basically measures flat with this method), but also exposes the worse treble quality than the HD800, as can be seen from the CSDs above. The banana pulp also has much slower attach/decay characteristics.

    I think that despite a more rough FR and resonances, distortion and off-axis issues what I really want is a bigger, more expensive widebander. This driver really doesn't come close to good headphones in technicalities. Bigger widebanders are probably faster sounding, more efficient, more resolving, more dynamic and just generally better sounding.

    Here's the FR with only one speaker playing:
    InEarOne.png
    The bass looks anemic now, but bass is generally mono information, so I think the upper graph is more accurate in this regard. What's interesting is the difference between 30° angle to the speakers (looking straight) and 0° angle (nose facing the speaker): This is essentially my version of the "BBC dip". I want a voice that's centered between the speakers played back by stereo speakers to sound the same as if it was played on a single mono speaker. Looking at the curve above** we would have to lower the region between 2 and 5kHz by about 3db or so to achieve this. The stuff below 1kHz is too much affected by general positional differences, so I wouldn't interpret too much into it. It's probably the same or at least very similar below 1kHz.

    Here's the step response with one speaker playing, looking straight ahead:
    IR.png
    Looks like you get a much better driver integration now, but I think that's just because the 2kHz dip is gone. This seems realistic to me.

    Note: Ignore the 50Hz hum issues. This is a problem with my mic pre.

    *Treble is often mixed less as mono information, so it's a better idea here to look at the 2nd graphs. Subjectively anything more than about 5db was too much. Still not a ton of air, but that's the price you have to pay with widebanders, even more so with real widebanders.

    **Actually we would have to look at the difference between stereo and mono, this graph:
    InEarStereoVsMono.png
    The other one is close enough.

    The bottom line here is that too much measurbation is ghey. Listen. Trust your ears. My graphs were biasing me into thinking that the treble rolloff that I heard wasn't such a big deal, but it is there. Pick drivers and speakers based on how they sound, not on how flat the graphs are. I actually think EQing for a flat (or B&K) response is wrong, since it will probably not sound flat.
     
    Last edited: Jan 14, 2017
  20. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    7,959
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    Not that much since I think you put the mics quite a bit out of the ear. Try moving the mics inside your ear a bit more and see the magical hump appear.

    Also, you proly are facing the speakers. Which means the mics may be 90 degrees relative to the tweeter. Which means roll-off.
     

Share This Page